add apikey

This commit is contained in:
2025-03-27 01:14:18 -03:00
parent 7021833476
commit 8118dfd403
14 changed files with 114 additions and 69 deletions

View File

@@ -8,7 +8,7 @@ from tqdm import tqdm
from boto3clients import dynamodb_client
elastic_client = Elasticsearch('http://127.0.0.1:9200')
files = (
jsonl_files = (
'test-orders.jsonl',
'test-users.jsonl',
'test-enrollments.jsonl',
@@ -16,7 +16,7 @@ files = (
)
def put_item(item: dict, table_name: str, *, dynamodb_client) -> bool:
def put_item(item: dict, table_name: str, /, dynamodb_client) -> bool:
try:
dynamodb_client.put_item(
TableName=table_name,
@@ -28,7 +28,7 @@ def put_item(item: dict, table_name: str, *, dynamodb_client) -> bool:
return True
def scan_table(table_name: str, *, dynamodb_client, **kwargs) -> Generator:
def scan_table(table_name: str, /, dynamodb_client, **kwargs) -> Generator:
try:
r = dynamodb_client.scan(TableName=table_name, **kwargs)
except Exception:
@@ -45,6 +45,31 @@ def scan_table(table_name: str, *, dynamodb_client, **kwargs) -> Generator:
)
class Elastic:
def __init__(self, client: Elasticsearch) -> None:
self.client = client
def index_item(
self,
id: str,
index: str,
doc: dict,
):
return self.client.index(
index=index,
id=id,
document=_serialize_python_type(doc),
)
def delete_index(self, index: str) -> bool:
try:
self.client.indices.delete(index=index)
except Exception:
return False
else:
return True
def _serialize_python_type(value: Any) -> Any:
if isinstance(value, dict):
return {k: _serialize_python_type(v) for k, v in value.items()}
@@ -58,53 +83,27 @@ def _serialize_python_type(value: Any) -> Any:
return value
def index_item(
id: str,
index: str,
doc: dict,
*,
elastic_client: Elasticsearch,
):
return elastic_client.index(
index=index,
id=id,
document=_serialize_python_type(doc),
)
def delete_index(index: str, *, elastic_client: Elasticsearch) -> bool:
try:
elastic_client.indices.delete(index=index)
except Exception:
return False
else:
return True
if __name__ == '__main__':
for file in tqdm(files, desc='Processing files'):
elastic = Elastic(elastic_client)
for file in tqdm(jsonl_files, desc='Processing files'):
with jsonl.readlines(f'seeds/{file}') as lines:
table_name = file.removesuffix('.jsonl')
for line in tqdm(lines, desc=f'Processing lines in {file}'):
put_item(line, table_name, dynamodb_client=dynamodb_client)
put_item(line, table_name, dynamodb_client)
for file in tqdm(files, desc='Scanning tables'):
for file in tqdm(jsonl_files, desc='Scanning tables'):
table_name = file.removesuffix('.jsonl')
delete_index(table_name, elastic_client=elastic_client)
elastic.delete_index(table_name)
for record in tqdm(
scan_table(
table_name,
dynamodb_client=dynamodb_client,
dynamodb_client,
FilterExpression='sk = :sk',
ExpressionAttributeValues={':sk': {'S': '0'}},
),
desc=f'Indexing {table_name}',
):
index_item(
id=record['id'],
index=table_name,
doc=record,
elastic_client=elastic_client,
)
elastic.index_item(id=record['id'], index=table_name, doc=record)